ramp-resources doesn't reuse unchanged compiled files

[jodal]

ramp-resources doesn't reuse unchanged compiled files, so buster-coffee causes all files to becompiled on every test run. We discussed this on IRC in August, but we never resolved this:

2012-08-15 15:54:32     jodal   cjno: compiling my set of coffee files takes
    ~1200ms every time I run `buster test`. where should I start looking to
    understand how to make buster only recompile files not already
    compiled and cached at the captured browsers? I assume somewhere
    in ramp-resources?
2012-08-15 16:51:10     jodal   cjno: so far I've figured out that my
    compiled js resources didn't have the etag set, so now I'm copying the
    etag from the original .coffee resource to the compiled .coffee.js
    resource
2012-08-15 16:51:49     jodal   doens't reduce the time spent compiling,
    though, but http://localhost:1111/resources now lists the .coffee.js
    files
2012-08-16 08:34:52     cjno    jodal: hmm, I have to look into that
2012-08-16 08:35:02     cjno    jodal: possibly this is because the
    resources change names?
2012-08-16 09:33:27     jodal   cjno: they have the same name every time,
    e.g. name from the file system + ".js"
2012-08-16 09:34:16     cjno    yeah, but the input file is named different
    from the output file, right?
2012-08-16 11:46:28     jodal   cjno: before buster-coffee runs, the
    resourceset got e.g. foo.coffee, and it's in the loadpath. after
    buster-coffee runs, the resourceset got foo.coffee and foo.coffee.js,
    but only foo.coffee.js in the loadpath. the etag of foo.coffee is copied
    onto foo.coffe.js so that the compiled resource has an etag that only
    changes when the source file changes.
2012-08-16 12:29:11     cjno    jodal: I just wonder if Buster sees that
    the *.js file is not compiled, and
2012-08-16 12:30:00     cjno    jodal: I will have to investigate tonight
2012-08-16 12:31:19     jodal   note that the last buster-coffee release
    doesn't set the etag. I changed that in git only yesterday

The etag fix in buster-coffee is still unreleased, but as mentioned, it doesn't seem to be enough to resolve this issue.

[cjohansen]

Hmm, was this a problem before or after I implemented the resource alternatives approach we discussed on web rebels?

[jodal]

This is with v0.6. I've never tried with master/v0.7. When did you implement the resource alternatives?

[cjohansen]

It's on master/0.7. Will ping you about it when I get it to a stage where you can play with it (the binaries are currently out of commission).

[jodal]

buster-coffee 0.1.4 released last week includes the ETag on compiled *.coffee.js files.

Together with Buster.JS 0.6.12 I still have 1-2 seconds of CoffeeScript compilation before every test run, even though the browser typically got N-1 up to date *.coffee.js files cached.

[dkl-ppi]

@jodal, has something changed with buster.js 0.7?

[jodal]

Not that I know of, but I don't really know, as I don't use buster-coffee myself any more.